The invention generally relates to a digital-to-analog converter (DAC), particularly for systems requiring an ultra-fast high resolution conversion of a digital input code into a corresponding analog output voltage. The DAC has an inherent low impedance bipolar output.
Conventional DACs comprise a switch matrix controlled by the digital input code, and an associated resistor network which is used for obtaining binary weighted partial currents or voltages for a further adding in a summing unit. The partial currents are obtained by current sources each coupled to the summing unit directly or via a separate switch. The accuracy of the current sources practically determines the accuracy of the DAC. Furthermore, the switch on-resistance, linearity, temperature coefficient, etc. are insignificant. An output amplifier is necessary for obtaining a voltage proportional to the input code.
The DACs employing the resistor network for obtaining binary weighted voltages must use switches having precisely matched characteristics as the on-resistance thereof is added to selected resistances of the resistor network. A constant voltage drop across every closed switch must be maintained. However, the circuit structure is fairly simple as the reference voltage is directly divided by the resistor network.
Several designs combine good and bad features of both techniques described above. For instance, commonly known quad current source approach is based on binary weighted current sources, in groups of four. Parts of the circuit are repetitive which makes easier a monolithic integration. For instance, three such quads provide a 12 bit resolution. The circuit still requires current sources operating with three different currents and two accurately trimmed resistor networks with low temperature coefficients.
The disadvantages of conventional DACs are many. A resistor network is required with precisely trimmed resistors having up to N different values for an N-bit DAC. The binary weighted currents or voltages must employ respectively precisely matched switches or transistors of the current sources. Specific serial and/or parallel connections of the actually resistive switches or the current sources are often necessary for matching respective parameters and their temperature coefficients. As a result the monotonicity is difficult to achieve and the long term stability is poor. The parasitic capacitances as well as switching delays of the analog switches cause high transient voltage spikes. Output amplifiers are employed as DACs with current output and often low output impedance dependent on input code are of little use. Highly accurate output amplifiers must then be employed by the user as the inaccuracies of the amplifier are not initially taken into account.